I bought a 100 meter Fathom ROV Tether (by the meter) hoping I would be able to do gigabit or at least 100BASE-TX speeds. I was disappointed that the first crimp on both end of the tether started with 10Mb/s. Thinking that maybe the tether I was given was over 100 meter, I begin cutting it meter by meter but each time only reaching a max speed of 10Mb/s. Only after cutting the tether by 20 meters or so was I able to get 100Mb/s. In the description of the tether, it states “These pairs are colored and arranged in the same fashion as a Cat5 networking cable including cross-talk resistance.” Cat5 network cable should be able to do at least 100Mb/s at 100 meters, not 10Mb/s.
I don’t know enough about ethernet cable standards or our cable construction to give a definitive response to this, so I’ve asked about it internally and will get back to you
That said, “colored and arranged in the same fashion as a Cat5 networking cable” at least to me doesn’t imply that it’s certified or rated to the Cat5 standard, just that it’s made in a similar way (so should hopefully have similar properties, but isn’t guaranteed to).
I think it would be useful if we could provide an official bandwidth rating/performance specification for our tethers (for both ethernet and powerline-based communication as relevant), so I’m also interested in what the response is.
I had a different interpretation to the wording of Cat5 networking cable part in the product’s description.
Cat5 networking needs a specific number of twists per meter in each twister pair so stating arranged in similar fashion provides the context that the way its twisted is the same as Cat5. Other than the twister pairs and the wire gauge, there is nothing special that makes Cat5 cabling unique. Either way, I hope we can come to a resolution with this since the advertising can be easily misled.
Ok, then we may need to revise the wording once we have a more definitive answer on what it’s intended to mean
I believe there are Cat5 cables with different wire gauges, and if it’s similar to other specifications Cat5 may also have requirements on the conductor material, resistance, separation, shielding, etc, or could be partially performance-based, and in either case “the same colours and twisting” may not imply full compliance.
As mentioned though, I haven’t looked into the specification in depth, or our cable construction, so it’s possible Cat5 is defined entirely by the colours and twist frequency, and it’s possible our cables are (or are not) expected to meet the Cat5 specification. We’ll have to see
As an initial response from one of our design engineers,
That’s a performance expectation though, and not a specification compliance statement. Hopefully our primary cable engineer will provide a more official response in the next couple of days.
Out of interest, how are you testing the bandwidth? Given cutting off several meters of tether apparently resulted in a sharp jump from 10Mb/s to 100Mb/s, is your testing process structured in floor-based levels (e.g. 98Mb/s is less than 100, so registers as the next level down, which is 10Mb/s)? From a level-based specification-compliance point of view that would technically be the correct approach for evaluation, but that would then be a poor metric for understanding practical/usable bandwidth.
I tested it by connecting the ethernet to a gigabit connection where the computer showed the max speed as 10Mb/s (in the ethernet properties settings for Windows 10) and doing internet speed test showed that it maxed out at 10. I also tested it on ArduSub Companion and running the network speed test also showed 10Mb/s. Its when the tether was about 60 meters or so that the ethernet properties starts detecting the ethernet connection as 100Mb/s and there I was able to do an internet speed test and a RPi network test which received 100Mb/s.
Hmm, fascinating. I wonder if ethernet itself has built-in level-based cutoffs, for performance consistency across systems / networks or something. That seems like a weird way to handle communication, but it’s not impossible.
If that is the case, I’d be curious what the achieved rates are through a system that’s able to report/present itself as 100Mb/s and then just gets as much data as possible through the connection. It’s possible the Fathom-X boards do something like that, although since that also uses a completely different communication medium through the tether (and generally only uses a single pair) it probably wouldn’t make for a particularly meaningful comparison.
Yes, the Fathom-X uses a different bandwidth to achieve the extra distance. We were going to use it if our tether length reached the max for ethernet but since it didn’t we didn’t see a reason to. Additionally, our setup uses multiple exploreHD cameras so we want the slightly higher bandwidth of 100Mb/s as oppose to 80Mb/s. Either way, I hope you guys can provide some resolution for us!
I had a similar problem some time ago - my first solution was to buy a special cable with CAT5 ethernet and 2 powercables in it. This cable is far from being “user friendly” and it became quite heavy too. In the end I just put it in the garage (300m of perfect neutral cable up to 300m water depth ). The second solution was a 2Wire ethernet range extender - which really works great and my blueview just needed 4 wires then: 2 ethernet, 2 power.
I appreciate your solution but we dont have enough space for an ethernet extender and its too late now since we already cut the tether
Also we really want 100Mb/s and most ethernet extender would reduce to 80Mb/s. Our plan was to stream multiple exploreHD cameras and reduce the amount we have to lower the bitrate by for the best quality.
Our manufacturer makes the tether with the same pairs and twists per meter as Cat-5, so in theory it could be Cat-5 compliant. That said, we haven’t tested whether it meets the Cat-5 performance specification, so can’t guarantee that it does.
Given that, I’ve asked that we do some performance testing over different tether lengths so that we can provide expected performance values, but we likely won’t have capacity to do that testing until a couple of months.
It’s perhaps worth noting that in our own use we only really send IP communication through our tethers via Fathom-X boards, and any extra tether pairs are then able to be used for alternative purposes. While it’s definitely possible to crimp on some ethernet connectors and use our 4-pair tether as a direct ethernet cable, it’s not a use-case we tend to operate or test with (so apologies that we don’t have more information readily available about it).
Given the HomePlug protocol is supposed to be able to support multiple devices on the same lines, you may be able to get additional bandwidth by using extra Fathom-X boards through the same wires. That said,
that’s not something I’ve tested (it may not work),
it could get expensive quite quickly,
it would take up quite a bit of enclosure space, and
it may need a network switch at the topside to combine the outputs (assuming your computer doesn’t have multiple ethernet ports)
While I find that idea interesting, it’s likely not worth hedging bets on for a project with an impending deadline.
I recently learned some things about this the hard way when trying to do POE over long cables - stranded vs. solid core wires makes a big difference! Solid core, 24 gauge wire will do 90+ mbps (hard to get the full 100) at lengths of 60m, but stranded wire, even lower gauge, won’t even connect! The RF magic that goes into ethernet is sensitive to construction, for sure!
100M is on the bleeding edge of the specification. Additionally, i’ll bet that spec is in free air, not in water, which is worst case for power. Just as you have to run larger conductors for in-ground power wiring, the same is true for comm cable. Especially true for high frequency comm cable as the signal travels through the diaelectric. The test here is link speed. Once the link speed gets the result you want, then you can do some performance testing to see if you get transmission errors. Your target errors should be zero.
If BR does some testing, make sure to test under water and coiled and not coiled (or as few coils as possible). My guess is that 100m underwater would be difficult to achieve. When you terminate the cable, there are spe4cifications for that as well. You should be OK for the spec if the twists continue all the way into the plastic shell. This is easiest to achieve if you use pass-through type cat5 connector ends. Cat 6 offers more protection, but makes for a much stiffer cable.
I’m curious what would occur if auto-negotiation is turned off, and 100Mbps set as the only option. It’s possible that would at least try to do faster speeds, although I don’t have enough experience with ethernet to know whether that would cause its own issues. It’s possible that configuration would need to be done at the topside as well, instead of just in the vehicle, but again that’s not something I’ve tried / have experience with.